This Small Business Innovation Research (SBIR) Phase I project proposes a bimanual robotic rehabilitation system, named BiArm, for stroke survivors who experience hemiparesis - weakness on one side of the body. The system is comprised of two special three-dimensional robotic arms designed to interact through forces with people over large reachable workspaces, reacting with feather-touch forces in addition to gravity- offsetting forces. The terminus of each of the robotic arms is attached either to the forearm or wrist of a subject via a cuff. Forces are then imparted by the robotic arm via the cuff in the full three-dimensional space surrounding the patient, allowing a range of full-motion functional activities that the patient can complete in a virtual environment generated by the robotic arms. Coordinated two-armed activities of daily living (ADL) can include lifting large virtual boxes onto virtual shelves, balancing virtual two-handle trays supporting an uneven array of weighty objects, and scrubbing virtual surfaces with both hands using a large virtual brush. Initially, it is hoped that this tool will enable clinical researchers in stroke rehabilitation to experiment with various modalities for bimanual interventions. If it can be shown that such a device improves the recovery from stroke beyond that of conventional techniques, then the business plan will ultimately be to provide an affordable bimanual device that can be used as a tool by rehabilitation clinicians. The proposed SBIR Phase I activities are divided into 3 aims. The first aim is to integrate two single-arm systems into a bimanual rehabilitation system (BiArm). This aim leverages a single-arm device (Proficio) just developed under a DARPA (Defense Advanced Research Projects Agency) grant. This single-arm is built for haptic-interactions with the hand, wrist, or forearm of people, depending on the endpoint attachment. The first aim will be to integrate two separate systems into a bimanual system in which there is rich interaction between both robot arms and both of the patient's arms. The second aim will be to generate software that enables the user to engage in functional unimanual and bimanual tasks seeing the 3-dimensional perspective representation of the task on a large-format computer monitor and feeling the virtual forces that would be expected in such activities. The third aim is to study the usability of such a system when used with hemiparetic patients to seek whether the normal arm positively affects the weak arm when the patient is focused on a bimanual task. Six (6) hemiparetic patients at Spaulding Rehabilitation Hospital in Boston, MA, will be selected for single-session interactions with the device, specifically performing two of th virtual ADLs. The system will be observed and evaluated for usability.